Projects | Lukas Pfannschmidt

Feature Relevance Analysis Tool

Thu, 02 May 2019 14:03:32 +0200

Feature selection is the task of finding relevant features used in a machine learning model. Often used for this task are models which produce a sparse subset of all input features by permitting the use of additional features (e.g. Lasso with L1 regularization). But these models are often tuned to filter out redundancies in the input set and produce only an unstable solution especially in the presence of higher dimensional data.

FRI calculates relevance bound values for all input features. These bounds give rise to intervals which we named ‘feature relevance intervals’ (FRI). A given interval symbolizes the allowed contribution each feature has, when it is allowed to be maximized and minimized independently from the others. This allows us to approximate the global solution instead of relying on the local solutions of the alternatives.

K-means OpenCL

Mon, 04 May 2015 18:24:17 +0200

In this project we used the K-means algorithm as an example to learn GPU computing using the OpenCL language.

Adverse Drug Effect Checker

Sun, 03 May 2015 15:54:25 +0200

One important task of medical practitioners is the prescription of drugs. Because drugs can act similarly in the human body they could exhibit harmful effects when used together. Its crucial for doctors to keep these adverse interactions in mind when writing prescriptions. Given the ever growing amount of data and more knowledge about these interactions this task is getting harder.

The goal of this project was a user oriented Android app to help doctors in their decision making. It allows the input of medication and patients gene defects to check in an integrated database for critical combinations. This project began as part of an undergrad course and is limited in scope and should be regarded as a prototype!

Screenshots:

Sublimator Controller

Sat, 03 May 2014 21:20:43 +0200

The overarching goal of this project was the development of a new machine used in biological experiments. Biologists often utilize mass spectrometry to analyse samples. To improve the analysis, it is necessary to use a homogeneous sample. One can use the process of sublimation to turn solid substances into gas. This gas then finally turns into a perfectly homogeneous solid again, when cooled down. To automate this process one can built a sublimaton machine.

Control Interface

In this project we used a Raspberry Pi microcomputer to control a heating element as well as vacuum pump to create a Sublimator. Our contribution specifically was the implementation of the control software which enabled preprogrammed heating schedules as well as logging features. The interface can be seen in Figure 1.

This project was a group effort. Also involved in the project were Jens Schulz and Dominik Gründing.

Example heating sequences

Efficient Edit Distance

Sat, 04 May 2013 19:00:09 +0200